Olg 2017 03 25 20 33 20 2

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

1

The Overlapping Generations Model (OLG)

In the OLG model, agents live for two periods. When young

they work and divide their labour income between consumption and

savings. When old they consume their savings. As the name of the

model implies, generations overlap so in each period there are

young agents saving and old agents consuming.

This model allows us to consider two issues, the consequences

of finite time horizons and the consequences of difference between

agents.

We begin with the agent's consumption saving decision.

Consumers born in generation t get labour income wt and consume cty

when young and cto = (1+rt+1)(wt-cty) when old.

Consumers maximize U(cty,cto). For example, they might

maximize log(cty) + (1/(1+d))log(cto) or slightly more generally

u(cty) + (1/(1+d))u(cto) for some function u. We assume that they

choose cty rationally so

1) u'(cty) - ((1+rt+1)/(1+d))u'(cto) = 0
For a fixed wage wt, different interest rates rt+1 can cause a

pattern of consumption like that shown in figure one or like that

shown in figure two. Where x = (cy(wt,rt+1),co(wt,rt+1)). An

increase in rt+1 must cause an increase in c to but may cause cty to

increase or to decrease.
2

In the simplest type of OLG model there are no long lived

assets. This means that there is no way to save so 1+r t+1 = 0 (see

Samuelson 1958 or Cass and Yaari 1968). In this case people

starve when they are old. This illustrates one unusual feature of

OLG models; the market outcome may be very bad. In fact the

market outcome may be Pareto inferior to some other feasible

outcome.

The grim world described above can clearly be improved if the

young give some fraction of their income to the old each period.

Everyone can be made better off by such a social security system.

Indeed, the population grows by a factor 1+n each generation,

everyone can get 1+n units of consumption good when old for each

unit of consumption good given away when young.

The optimal steady state is achieved when each of the young

contributes x units to the scheme and each of the old get x(1+n)

units and

2) u'(wt-x) - ((1+n)/(1+d))u'(x(1+n)) = 0.
3

This is what a social planner who cared only about the steady

state level of happiness would require people to do.

Such a desirable outcome can be achieved without a social

security system. Let's say that there is one unit of

intrinsically worthless asset, say pieces of paper with pictures

printed on them (money). If these cannot be counterfeited, they

may be valuable in equilibrium in an OLG model. Each generation

of young may buy the money of the old because they know that the

new young will buy the money form them when they are old.

If pt is the amount of money required to buy one unit of

goods at time t then the return to holding money is p t/pt+1, for

each unit of consumption good given up at t one gets p t/pt+1 units

of good when old. In other words 1+rt+1 = pt/pt+1.

In the case of wt fixed at w, total production in time t is

Ntw where Nt is the number of young born at t. Since the good can

not be stored, total consumption equals total production so

2) ct-1oNt-1 + ctyNt = wNt


This means that the value of the money 1/pt is given by

3) 1/pt = Nt-1ct-1o = Nt(w - cty)

A little algebra tells us that

4) cty = w - (1/(1+n))ct-1o.

This and the first order condition of the young

5) u'(cty) - pt/((1+d)pt+1)u'(cto) = 0

mean that we can calculate the equilibrium pt+1 from pt. This
4

calculation can be done graphically where cty is calcualted from

ct-1o by drawing the horizontal line to the graph of 4) the line

through (w,0) with slope -1+n. Then cto = 1/(Ntpt+1 can be

calculated using equation 5 by drawing the vertical line to the

offer curve, the graph of equation 5.

The optimal steady state defined by 2 can always be an

equilibrium of the overlapping generations economy with money.

The value of the money at t has to equal xN t for x as defined in

equation 2. This means that money has to become more and more

valuable as the price of the good declines by a factor 1/(1+n)

each period. This is an equilibrium no matter what the shape of

the offer curve.

The optimal steady state is not the only possible equilibrium

of the overlapping generations economy with money. Figure 3 shows

an inflationary equilibrium in which the stock of money becomes

less an less valuable compared to wNt. This implies that those

unfortunate enough to be born at a high t starve when they are old


which is very sad.
5
6

In contrast if the offer curve is as in figure 4 a deterministic

cycle is possible in which e.g. even numbered generations consume

a large amount both when young and when old, but odd numbered

generations consume little when young and when old which is less

sad but not fair.


7

As figure 5 shows the optimal steady state can be a stable

equilibrium so the economy ends up there even if it started

somewhere with p different from the optimal p but not to far from

it.

These figures also illustrate another odd feature of overlapping

generations economies, the equilibrium outcome is not unique.

Equations 4 and 5 allow us to calculate p2, p3 &c for many

different values of p1. This means that knowledge of the tastes

and the technology does not enable us to say if the market outcome

will be Pareto optimal or very very bad.


In the case of the offer curve in figure 5 there are also
equilibria in which something truly strange happens. In period 3,
8

the Walrasian auctioneer flips a coin. If it comes up heads


(testa so t) then p3 is high so the old have low consumption c2o
is low and c3y is high and vice versa if the coin comes up tails
(croce so c). When 2nd generation agents are young, they know this
will happen and they choose consumption under uncertainty. The in
each case (testa or croce) there is a series of prices which makes
markets clear. Thus the equilibrium can be stochastic even if
tastes and technology are not. In figure 6 “etc”. is et cetera,
that is I claim that I could show the continuation of the testa
equilibrium (if I could draw better).

You might also like